Wuxin Yue

Developed a Retrieval-Augmented Generation (RAG) pipeline integrating a Large Language Model (LLM) with a vector database (e.g., FAISS) to provide accurate, context-aware answers for domain-specific queries. Designed and optimized Prompt Engineering strategies to improve LLM output quality and reduce hallucination rates by 20%. Built API endpoints using FastAPI for seamless integration into business workflows, enabling real-time query handling. Implemented automated evaluation scripts to benchmark retrieval accuracy and generation quality across multiple datasets